What Is a Marketing Funnel and How Does It Actually Work?

Updated September 2026 · Written and maintained by the Progression Agency strategy team

A marketing funnel is a diagnostic tool, not a description of how people behave. Real buying is non-linear, involves several people and is only partly visible — and the model is still useful, because it shows where people stop. This covers what each stage means, how to find the one that is actually costing you, what fixes each one, and why the instinct to buy more traffic is usually addressing the wrong end.

The short answerThe stage with the worst conversion ratio is not automatically the one to fix. The one to fix is where a realistic improvement produces the largest absolute gain — which requires doing the arithmetic rather than reacting to the worst-looking percentage. And verify the tracking before believing any of it: a large share of funnel analysis is performed on numbers that were never reliable.

Progression Agency is based in New York City and works with clients across the United States and worldwide. Funnel models, stage names and measurement approaches vary widely between organizations; what matters is internal consistency and whether the stages correspond to something you can actually observe.

Marketing funnels in short — Spending on awareness when the problem is conversion is the most expensive mistake in this area, and it is extremely common.

What is a funnel in marketing?

A model describing how somebody moves from not knowing you exist to buying — used to work out where people are dropping out and what to fix.

It is a diagnostic tool rather than a description of how people actually behave. Real buying is messy, non-linear and involves several people; the funnel is useful precisely because it simplifies that into something you can measure.

The stages, and what each one means — Most funnel diagrams stop at purchase, which is the stage after which the money actually gets made.

What are the stages, and what does each actually mean?

Awareness, interest, consideration, intent, evaluation and purchase — though the number of stages matters far less than being able to measure movement between them.

Any model works if the stages correspond to something you can observe. A six-stage model you cannot measure is worse than a three-stage one you can.

Awareness

They learn the problem or the category exists. Frequently they do not know your name at this point and are not looking for you.

Interest

They start seeking information. This is where search demand first becomes visible and where content earns its keep.

Consideration

They are comparing approaches or providers. Comparison content and reviews carry most of the weight here.

Intent

They have decided to act and are choosing between specific options. This is where paid search captures most of its value.

Evaluation

The practical checks: price, terms, references, whether you answer the phone. Many sales are lost here for entirely non-marketing reasons.

Purchase

The transaction. In B2B it involves people who never appeared in any of your analytics.

Retention and advocacy

Everything after the sale, which is where profitability actually lives and which most funnel diagrams omit entirely.

Why does the funnel model get criticized?

Because real buying is not linear, several people are involved, and people enter partway through — all of which are true and none of which make the model useless.

The honest position is that a funnel is a measurement scaffold rather than a theory of human behavior. Used to find where people drop out, it works. Used as a description of a journey, it misleads.

Fair criticisms of the funnel, and what to do about them
Criticism	Is it fair?	What to do
Buying is not linear	Yes	Measure stages, not a journey
People enter partway through	Yes	Do not assume top-of-funnel entry
Several people are involved in B2B	Yes	Map the buying group, not one person
It ignores retention	Yes, as usually drawn	Extend it past purchase
It implies you control the process	Yes	You influence, not control
Attribution between stages is imperfect	Yes	Use it directionally
It is too simple	That is the point	Simplicity is what makes it measurable
Everybody uses a different model	True and unimportant	Consistency internally is what matters

The most useful reframe: stop asking whether the funnel is accurate and start asking whether you can see where people stop. If you can, the model is doing its job.

How do you find where people are actually dropping out?

By measuring the ratio between each observable stage, then comparing those ratios against what the business can plausibly change.

The stage with the worst ratio is not always the one to fix. The one to fix is where a realistic improvement produces the largest absolute gain.

Count each stage — Diagnose. Whatever you can observe..

Calculate the ratios — Diagnose. The ratio is the signal..

Check it is fixable — Diagnose. Some constraints are not..

Model absolute gain — Diagnose. Not the worst percentage..

Verify the tracking — Diagnose. Before believing anything..

Split branded traffic — Diagnose. It flatters everything below..

Count what enters each stage

Sessions, inquiries, qualified opportunities, proposals, closed deals. Whatever you can actually observe.

Calculate the ratios between them

The conversion from each stage to the next is the diagnostic number, not the absolute counts.

Find the worst ratio, then check it is fixable

A poor ratio caused by an unfixable constraint is not where to spend.

Model the absolute gain, not the percentage

Doubling a tiny number at the bottom frequently beats a small percentage gain at the top, and frequently does not. Do the arithmetic.

Check the tracking before believing any of it

A large share of funnel analysis is performed on numbers that were never reliable.

Separate branded from non-branded at the top

Branded traffic entering the funnel flatters every stage below it.

Which stage do most businesses actually have a problem at?

More often than not, further down than they think — inquiry handling and follow-up rather than traffic.

The instinct when sales are down is to buy more traffic. Frequently the traffic is adequate and the inquiries arriving are being answered slowly or not at all.

Where businesses most often lose people — Relative frequency, not measured values. The bottom row is where budget goes first and the top row is where the problem usually is.

What actually fixes each stage?

Different work entirely — which is why diagnosing the stage matters more than the sophistication of any individual tactic.

Spending on awareness when the problem is conversion is the most expensive mistake in this whole area, and it is extremely common.

What fixes each stage
Stage	Symptom	What actually fixes it
Awareness	Nobody has heard of you	Reach: search, paid, PR, partnerships
Interest	Traffic that leaves immediately	Content that matches what was searched
Consideration	They compare and pick somebody else	Comparison content, proof, reviews
Intent	They search and choose a competitor	Paid search, landing pages, offer clarity
Evaluation	Proposals that go quiet	Response speed, references, pricing clarity
Purchase	Deals stall at the end	Process friction, terms, decision-maker access
Retention	Customers leave after one purchase	Onboarding, service, communication
Advocacy	No referrals	Ask; most businesses simply never do

The bottom row is the cheapest line in the table and the one most often left blank. Referrals arrive when somebody asks for them, and most businesses have no process that ever does.

How does the funnel differ between B2B and consumer?

B2B funnels are longer, involve several people, and include stages that happen entirely offline and invisibly.

The practical consequence is that B2B funnel measurement is always partial, and treating the visible part as the whole picture produces confident wrong conclusions.

B2B and consumer funnels compared
Consideration	B2B	Consumer
Length	Months to years	Minutes to weeks
People involved	Several, with different concerns	Usually one
Visible in analytics	Partially	Mostly
Deciding factor	Risk reduction	Value or desire
Role of content	Substantial, throughout	Concentrated near purchase
Role of sales	Central	Frequently absent
Where deals are lost	Internal approval	Checkout and price
Measurement quality	Partial and lagging	Fairly complete

The ‘people involved’ row is what defeats naive B2B funnel analysis: the person who searched is frequently not the person who decides, and the person who decides may never visit the site at all.

What is the difference between a funnel and a customer journey?

A funnel counts people at stages. A journey describes what an individual experiences. They answer different questions and are frequently confused.

Use a funnel to find where volume is lost. Use a journey map to understand why, and what the experience actually feels like at that point.

Do you need marketing automation to run a funnel?

No. You need to be able to count people at each stage, which a spreadsheet can do for most businesses below a certain size.

Buying automation before knowing which stage is broken is a common and expensive sequencing error: the tool then automates a process nobody has diagnosed.

Which stage should you actually fix? — The top-right cluster is nearly free and rarely done; the bottom-left is expensive and usually the first thing proposed.

A spreadsheet — Tools. Enough for most businesses..

Analytics, verified — Tools. Tested by submitting a form..

A CRM that is used — Tools. An unused one measures nothing..

Call tracking — Tools. Where phone inquiries matter..

Automation, later — Tools. After the diagnosis, not before..

One agreed definition — Tools. Of what each stage means..

How do you measure a funnel honestly?

By verifying the tracking first, separating branded from non-branded, accepting that attribution is imperfect, and using the numbers directionally.

Anybody presenting exact attribution across a multi-touch funnel is presenting a model, not a measurement. That is fine as long as everybody knows which it is.

What are the common mistakes?

Fixing the wrong stage, believing unverified tracking, ignoring everything after purchase, and treating the model as a description of behavior.

All four produce confident decisions built on the wrong foundation, which is worse than having no model at all.

How does a funnel change when buyers research with AI answer engines?

The interest and consideration stages increasingly happen somewhere you cannot see — inside an answer rather than on your site.

That does not break the model. It moves more of the early funnel out of your analytics, which makes the stages you can still observe more important and makes being citable more valuable.

Where each funnel stage is becoming harder to observe
Stage	Traditionally visible as	Increasingly happening
Awareness	Impressions, brand search	Inside an answer, unlogged
Interest	Site sessions, content reads	Summarized by an assistant
Consideration	Comparison page visits	In a generated comparison
Intent	Search clicks	Still mostly visible
Evaluation	Pricing page, contact	Still visible
Purchase	Conversion	Visible
Retention	Repeat orders	Visible
Advocacy	Referrals	Largely invisible either way

The practical response is not to try to recover the lost visibility. It is to make sure the stages you can still measure are measured properly, and to be the source those answers cite.

Not sure which stage is actually costing you?

We work with clients across the United States and worldwide, and the first thing we do is find where people actually stop — which, more often than anybody expects, is after the inquiry arrives rather than before it.

Talk to Progression Agency

Getting found in search

AI, AEO and what is changing

Paid media and lead generation

Websites and design

Choosing and working with an agency

Social, content and brand

By industry and by situation

Frequently asked questions

What does mkt funnel mean?

It is shorthand for marketing funnel. Mkt funnel appears in internal documents and job posts as an abbreviation, and it describes the same model: prospects segmented by how close they are to buying, with a message and a measurable step at each stage.

What funnel model should a small business actually use?

The simplest one that maps to your real sales steps. Asking what funnel to adopt matters less than whether the stages correspond to things that actually happen — an imported five-stage model that does not match your process produces reporting nobody trusts.

How do funnels work in practice?

They segment an audience by how close it is to buying. How do funnels work operationally: each stage has an entry definition, a message suited to that state, and a measurable step to the next. Without those three the funnel is a diagram rather than a system.

How does funnel work when the sale is not linear?

It stops describing a journey and starts describing a state. How does funnel work for considered B2B purchases where buyers loop back and forth: treat the stages as states a contact can occupy and re-enter, not as a one-way path, and the reporting stays honest.

What are funnels used for beyond marketing reporting?

Forecasting, resource allocation and diagnosing where deals stall. What are funnels used for in a well-run business is identifying the stage with the worst conversion, because that is where an hour of work returns most — which the top-line conversion rate hides.

What is a communication funnel?

A model of how a message moves from broad reach to individual response. A communication funnel describes attention narrowing through awareness, comprehension and action, which is useful for campaign planning even where no purchase occurs — public health and internal communications both use it.

How does customer funnel marketing differ from the sales funnel?

It covers what happens after purchase as well as before. Customer funnel marketing extends past the sale into onboarding, retention and advocacy, which matters because acquiring a replacement customer costs far more than keeping the current one — a fact the classic funnel’s shape obscures.

Where do you find reliable funnel information?

From your own numbers rather than from a model. Funnel information that helps is the ratio between stages you can actually observe in your business, not a diagram of stages you cannot measure.

What is a funnel in marketing?

A marketing funnel is a model describing how somebody moves from not knowing you exist to buying, used to work out where people drop out and what to fix.

Is it a description of how people really behave?

No. It is a measurement scaffold. Real buying is non-linear, involves several people and is only partly visible — the model is useful because it simplifies that into something measurable.

What are the stages?

Commonly awareness, interest, consideration, intent, evaluation and purchase — plus retention and advocacy, which most diagrams omit.

How many stages should a funnel have?

However many correspond to something you can actually observe. A six-stage model you cannot measure is worse than a three-stage one you can.

Is the funnel model outdated?

The criticisms are fair and the tool is still useful. Stop asking whether it describes a journey and start asking whether it shows you where people stop.

How do I find where people drop out?

Count what enters each observable stage, calculate the ratio between them, and look at where the ratio is worst — then check whether that stage is fixable.

Should I always fix the worst ratio?

No. Fix the stage where a realistic improvement produces the largest absolute gain, which requires doing the arithmetic rather than reacting to a percentage.

Where is the problem usually?

Further down than people think. Inquiry handling and follow-up far more often than a shortage of traffic.

Why is that so commonly missed?

Because response speed is invisible in advertising reporting, so a lead that arrived and was answered a day later shows up as the channel failing.

What fixes the awareness stage?

Reach: search visibility, paid media, PR and partnerships. It is also the stage most often bought when the problem is elsewhere.

What fixes the consideration stage?

Comparison content, proof, references and reviews — anything that gives somebody a reason to choose you over an alternative they are also looking at.

What fixes the evaluation stage?

Response speed, clear pricing, available references, and removing friction from the final steps. Many losses here have nothing to do with marketing.

What about after purchase?

Retention and advocacy are where profitability lives, and most funnel diagrams stop before them.

Why do so few businesses get referrals?

Because nobody asks. It is the cheapest line in the whole model and the one most often left blank.

How is a B2B funnel different?

Longer, involving several people with different concerns, and only partly visible — significant stages happen offline and never appear in analytics.

Why does that defeat naive B2B analysis?

Because the person who searched is frequently not the person who decides, and the decision-maker may never visit the site at all.

What is the difference between a funnel and a customer journey?

A funnel counts people at stages; a journey describes what an individual experiences. Use the funnel to find where volume is lost and the journey to understand why.

Do I need marketing automation?

No. You need to count people at each stage, which a spreadsheet handles for most businesses below a certain size.

When is automation the right purchase?

After you know which stage is broken. Buying it first automates a process nobody has diagnosed.

How do I measure a funnel honestly?

Verify the tracking first, separate branded from non-branded, accept that attribution is imperfect, and use the numbers directionally.

Why separate branded traffic?

Because branded searches come from people who already know you, and counting them at the top flatters every stage below.

Can attribution be exact?

No. Anybody presenting exact multi-touch attribution is presenting a model rather than a measurement — which is fine, as long as everybody knows which it is.

What if my tracking is unreliable?

Fix that first. A large share of funnel analysis is performed on numbers that were never reliable, and every conclusion drawn from them inherits the error.

What are the most common funnel mistakes?

Fixing the wrong stage, believing unverified tracking, ignoring everything after purchase, and treating the model as a description of behavior.

Is a simple funnel good enough?

Usually better than a complex one. Simplicity is what makes it measurable, and measurability is the entire point.

Should everybody in the company use the same model?

Internally, yes. It matters far more that your stages mean the same thing to everybody than that they match anybody else’s model.

Where should I start if I have nothing?

Count inquiries, count qualified opportunities, count sales. Three stages you can observe beat six you cannot.

What is the single cheapest improvement available?

Answering inquiries faster. It costs nothing, is invisible in reporting, and is where a large share of generated demand is lost.

Get a free marketing proposal

Tell us what you are trying to grow and we will come back with a plan, not a pitch deck. Same-day reply on weekdays.